Microtubule minus-end anchorage at centrosomal and non-centrosomal sites: the role of ninein.

The novel concept of a centrosomal anchoring complex, which is distinct from the gamma-tubulin nucleating complex, has previously been proposed following studies on cochlear epithelial cells. In this investigation we present evidence from two different cell systems which suggests that the centrosomal protein ninein is a strong candidate for the proposed anchoring complex. Ninein has recently been observed in cultured fibroblast cells to localise primarily to the post-mitotic mother centriole, which is the focus for a classic radial microtubule array. We show here by immunoelectron microscopical analyses of centrosomes from mouse L929 cells that ninein concentrates at the appendages surrounding the mother centriole and at the microtubule minus-ends. We further show that localisation of ninein in the cochlear supporting epithelial cells, where the vast majority of the microtubule minus-ends are associated with apical non-centrosomal sites, suggests that it is not directly involved in microtubule nucleation. Ninein seems to play an important role in the positioning and anchorage of the microtubule minus-ends in these epithelial cells. Evidence is presented which suggests that ninein is released from the centrosome, translocated with the microtubules, and is responsible for the anchorage of microtubule minus-ends to the apical sites. We propose that ninein is a non-nucleating microtubule minus-end associated protein which may have a dual role as a minus-end capping and anchoring protein.

Microtubule release and capture in epithelial cells.

Many differentiated cells including polarised epithelial cells display a non-radial, apico-basal microtubule array. In some cells the centrosome disassembles and new nucleating sites are created at more appropriate locations. In others the centrosome remains, but relatively few microtubules radiate from it's immediate environs. Instead, the majority of the microtubule minus-ends are associated with apical cell surface sites. Centrosomal microtubule release and capture is evidently a mechanism exploited by some polarised epithelial cells as a means of producing non-centrosomal, apico-basal microtubule arrays. This involves microtubule nucleation at the centrosome, release and subsequent translocation and capture at the apical sites. Two functionally distinct centrosomal complexes dedicated to the control of microtubule nucleation and anchorage have been suggested to be essential and universal features of all centrosomes. The centrosomal proteins ninein and R2 are potential microtubule anchoring proteins and their discovery has exciting implications for centrosomal organisation and microtubule positioning in cells.

Keratin filament deployment and cytoskeletal networking in a sensory epithelium that vibrates during hearing.

The intricate and spatially precise ways in which keratin intermediate filaments are deployed in certain cochlear epithelial cells, called supporting cells, suggests that these filaments make a micromechanically important contribution to the functional design of the guinea pig organ of Corti. Filament arrays that include keratins 8, 18, and 19 are confined mainly to regions close to the ends of large transcellular microtubule bundles in supporting cells. These cells and their microtubule bundles link sensory hair cells to a specialized basement membrane that vibrates during hearing. The keratin filament arrays apparently help anchor the ends of the microtubule bundles to cell surfaces. Filaments are concentrated at the apices and bases of most cells that contact hair cells. Substantial arrays of adherens junctions link the apices of these cells. Hence, keratin filaments may contribute to a cytoskeletal network that distributes mechanical forces from cell to cell and that coordinates the displacement of neighboring hair cells. However, high concentrations of keratin filaments have not been detected at the apices of one of the supporting cell types, which apparently has a mechanical role that is different from that of the others. Transmission electron microscopy has revealed previously undescribed filament networks at all the locations where the binding of antibodies to keratins is most marked. There is evidence that intercellular linkage of the keratin networks via their association with actin-containing meshworks and adherens junctions is more extensive than linkage provided by desmosomes.

Nucleation and capture of large cell surface-associated microtubule arrays that are not located near centrosomes in certain cochlear epithelial cells.

This report deals with the as yet undetermined issue of whether cell-surface associated microtubules in certain cochlear epithelial cells are centrosomally nucleated and subsequently migrate to microtubule-capturing sites located at the surface regions in question. Alternatively, the cells may possess additional nucleating sites which are noncentrosomal and surface-associated. These alternative possibilities have been investigated for highly polarised epithelial cells called supporting cells in the mouse and guinea pig organ of Corti using antibodies to pericentrin and gamma-tubulin. There is substantial evidence that both proteins are essential components of microtubule-nucleating sites in cells generally. Each mature supporting cell possesses a large microtubule array that is remotely located with respect to its centrosome (more than 10 microns away). The antibodies bind to a cell's centrosome. No binding has been detected at 2 other microtubule-organising centres that are associated with the ends of the centrosomally-remote microtubule array while it is being constructed. Such arrays include thousands of microtubules in some of the cell types that have been examined. If all a cell's microtubules are nucleated by its centrosome then the findings reported above imply that microtubules escape from the centrosomal nucleating site and migrate to a new location. Furthermore capture of the plus and minus ends of the errant microtubules is taking place because both ends of a centrosomally-remote microtubule array are attached to sites that are precisely positioned at certain cell surface locations. Minus ends are locating targets with an exactitude comparable to that which has been demonstrated for plus ends in certain cell types. These cells apparently operate a single control centre strategy for microtubule nucleation that is complemented by precise positioning of plus and minus end-capturing sites at the cell surface.

Centrosomal deployment of gamma-tubulin and pericentrin: evidence for a microtubule-nucleating domain and a minus-end docking domain in certain mouse epithelial cells.

This report provides evidence for two functionally and spatially distinct centrosomal domains in certain mouse cochlear epithelial cells. The vast majority of microtubules elongate from sites associated with the apical cell surface in these cells rather than from pericentriolar material surrounding the immediate environs of their apically situate centrioles. The distribution of gamma-tubulin and pericentrin at cell apices has been examined while microtubule nucleation is progressing because these centrosomal proteins are believed to be essential for microtubule nucleation. Antibodies to both proteins bind to pericentriolar regions but no binding has been detected at the apical cell surface-associated sites where the ends of thousands of recently nucleated microtubules are concentrated. Sparse transient microtubule populations can be detected between pericentriolar regions and surface sites while microtubule assembly advances. A procedure apparently operates in which the pericentriolar region functions as a microtubule-nucleating domain and the cell surface-associated sites operate as docking domains which capture the minus ends of microtubules that migrate to them shortly after nucleation. Docking domains may include some components of the pericentriolar material that have been relocated at the cell apex. A docking element hypothesis for centrosomal control of minus end positioning and dynamics in animal cells generally is proposed. This investigation has also shown that the concentration of gamma-tubulin and pericentrin around centrioles differs spatially and quantitatively in ways that are characteristic for the four cell types studied. Some of these characteristics can be related to differences in control of microtubule number and positioning.

Formation of two microtubule-nucleating sites which perform differently during centrosomal reorganization in a mouse cochlear epithelial cell.

This report provides evidence for the formation of a cell surface-associated centrosome with two spatially discrete microtubule-nucleating sites that perform differently; the minus ends of microtubules remain anchored to one site but escape from the other. Centrosomal reorganization in the cells in question, outer pillar cells of the organ of Corti, indicates that its pericentriolar material becomes intimately associated with the plasma membrane at the two nucleating sites. Two large microtubules bundles assemble in each cell. A beam which includes about 1,300 microtubules spans most of the cell apex. It is positioned at right angles to a pillar with about 4,500 microtubules which is oriented parallel to the cell's longitudinal axis. The beam's microtubules elongate from, and remain attached to, a centrosomal region with two centrioles which acts as a microtubule-nucleating site. However, the elongating microtubules do not radiate from the immediate vicinity of the centrioles. During beam assembly, the minus ends of the microtubules are concentrated together close to the plasma membrane (less than 0.2 micron away in many cases) at a site which is located to one side of the cell apex. High concentrations of the pillar's microtubules elongating from one particular site have not been detected. Analyses of pillar assembly indicate that the following sequence of events occurs. Pillar microtubules elongate from an apical cell surface-associated nucleating site, which becomes more distantly separated from the centriolar locality as cell morphogenesis progresses. Microtubules do not accumulate at this apical nucleating site because they escape from it. They migrate down to lower levels in the cell where the mature bundle is finally situated and their plus ends are captured at the cell base.

Three microtubule-organizing centres collaborate in a mouse cochlear epithelial cell during supracellularly coordinated control of microtubule positioning.

Large cell surface-associated microtubule bundles that include about 3,000 microtubules assemble in certain epithelial cells called inner pillar cells in the mouse organ of Corti. Microtubule-organizing centres (MTOCs) at both ends and near the middle of each cell act in concert during control of microtubule positioning. In addition, the three cell surface-associated microtubule-organizing centres are involved in coordinating the connection of bundle microtubules to cytoskeletal components in neighbouring cells and to a basement membrane. The precisely defined locations of the three MTOCs specify the cell surface regions where microtubule ends will finally be anchored. The MTOCs are modified as anchorage proceeds. Substantial fibrous meshworks assemble at the surface sites occupied by the MTOCs and link microtubule ends to cell junctions. This procedure also connects the microtubule bundle to cytoskeletal arrays in neighbouring cells at two of the MTOC sites, and to the basilar membrane (a substantial basement membrane) in the case of the third site. A fourth meshwork that is not positioned at a major MTOC site is involved in connecting one side of the microtubule bundle to the cytoskeletons of two other cell neighbours. The term surfoskelosome is suggested for such concentrations of specialized cytoskeletal materials and junctions at cell surface anchorages for cytoskeletal arrays. The large microtubule bundle in each cell is composed of two closely aligned microtubule arrays. Bundle assembly begins with nucleation of microtubules by a centrosomal MTOC that is attached to the apical cell surface. These microtubules elongate downwards and the plus ends of many of them are apparently captured by a basal MTOC that is attached to the plasma membrane at the bottom of the cell. In the lower portion of the cell, the microtubule bundle also includes a basal array of microtubules but these elongate in the opposite direction. This investigation provides evidence that they extend upwards from the basal MTOC to be captured by a medial MTOC which is attached to the plasma membrane and situated near the mid-level of the cell. However, there are substantial indications that the basal array's microtubules are also nucleated by the apically situated centrosomal MTOC, but escape from it, and are translocated downwards for capture of their plus ends by the basal MTOC. If this is the case, then these microtubules continue to elongate after translocation and extend back up to the medial MTOC, which captures their minus ends.

Reorganization of the centrosome and associated microtubules during the morphogenesis of a mouse cochlear epithelial cell.

Reorganization of centrosomal microtubule-organizing centres and the minus ends of microtubules occurs as the centrosomal ends of large microtubule bundles are repositioned and anchored to cell junctions in certain epithelial cells called inner pillar cells in the mouse organ of Corti. The microtubule bundle that assembles in each cell consists of two distinct microtubule arrays that run closely alongside each other. Both arrays are attached to the cell surface at their upper and lower ends. One of the arrays spans the entire length of a cell but the other is confined to its lower portion. Initially, about 3,000 microtubules elongate downwards from an apically situated centrosome in each cell. Subsequently, the minus ends of these microtubules, and the centrosome and its two centrioles, migrate for about 12 microns to the tip of a laterally directed projection. Then, a meshwork of dense material accumulates to link microtubule minus ends and the centrosome to cell junctions at the tip of the projection. Pericentriolar satellite bodies, which form after the initial burst of microtubule nucleation, may represent a condensed and inactive concentration of microtubule-nucleating elements. Surprisingly, as a cell matures, about 2,000 microtubules are eliminated from the centrosomal end of the microtubule bundle. However, about 2,000 microtubules are added to the basal portion of each bundle at levels that are remote with respect to the location of the centrosome. Possibly, these microtubules have escaped from the centrosome. If this is the case, then both the plus and minus ends of most of the errant microtubules are captured by sites at the cell surface where the ends are finally anchored. Alternatively, each cell possesses at least one other major microtubule-nucleating site (which does not possess centrioles) in addition to its centrosome.

Multiple plasma membrane-associated MTOC systems in the acentrosomal cone cells of Drosophila ommatidia.

Multiple plasma membrane-associated microtubule-organizing centers operate in the cone cells of Drosophila ommatidia. A transcellular array of about 250 microtubules assembles in each cone cell during late pupal ommatidial morphogenesis. While these arrays are assembling, cone cells do not possess conventional centriole-containing centrosomal microtubule-organizing centers. The microtubules are associated with plasma membrane-associated plaques at both the apical and basal surfaces. During assembly of the arrays there is a progressive decrease in the number of microtubules/cell cross section at successively lower levels in each cell which is indicative of apicobasal microtubule elongation. In this respect, assembly of the arrays closely resembles that of transcellular 15 protofilament microtubules in late pupal wing cells (Mogensen et al. J. Cell Biol. 108, 1445-1452 (1989)). However, cone cell microtubules are almost certainly of the 13 protofilament variety as found in eukaryotic cells generally. We suggest that plasma membrane-associated microtubule-organizing centers are widely employed in polarized epithelia in Drosophila during late pupal morphogenesis.

Microtubule rearrangement and bending during assembly of large curved microtubule bundles in mouse cochlear epithelial cells.

Mature inner pillar cells in the mammalian organ of Corti are curved through about 60 degrees, where they arch over adjacent epithelial cells and the apex of an intercellular space called the tunnel of Corti. This report deals with changes in microtubule organization that are associated with cell bending and tunnel formation during morphogenesis of the mouse organ of Corti. A large bundle of up to 3,000 microtubules assembles in each inner pillar cell. Microtubule rearrangement occurs about 5 days after bundle assembly begins. The lumen of each initially straight hollow tube-shaped microtubule bundle is occluded as the bundle becomes more compact and elliptical in cross section. This event anticipates the once-only bending which subsequently occurs between particular levels (about 9-19 microns) below the top of a bundle as it curves into its final shape about 2 days later. Microtubule rearrangement presumably facilitates bending which is effected in the plane of least mechanical resistance parallel to the short axis of a bundle's elliptical cross-sectional profile. Precocious bending of bundles has been induced about 1.5 days in advance of the natural event. Abnormal positioning of these prematurely curved bundles indicates that bending is effected by a contractile mechanism located within bundles rather than being a response to externally applied forces. The potential importance of such microtubule-associated contractions for active modulation of the vibratory response in the cochlea during hearing is considered.

A cell surface-associated centrosomal layer of microtubule-organizing material in the inner pillar cell of the mouse cochlea.

This investigation provides evidence that pericentriolar material is divorced from the immediate vicinities of centrioles and becomes functionally associated with the plasmalemma during the differentiation of a mammalian cell type. Such events occur prior to the assembly of large transcellular microtubule bundles in columnar epithelial cells called inner pillar cells in the mouse organ of Corti. The microtubules do not radiate from a typical centrosome and its centrioles. They elongate from a microtubule-organizing centre (MTOC), which is deployed as a subapical cell surface-associated layer in each cell. Most of the dense material of this layer, and the tops of most of the microtubules, are initially concentrated around the sides of a cell about 1 microns below its apical surface. In addition, a pair of centrioles is located above the layer, which acts as if it is a pericellular concentration of the pericentriolar material of a modified centrosome. Although microtubule nucleation takes place in a centrosome-like region, 13 protofilament fidelity is not exercised. Most of the microtubules have 15 protofilaments. Microtubule assembly progresses in these cells after the organ of Corti has been isolated for in vitro culture. However, large numbers of microtubules elongate from pericentriolar material juxtaposed against the centrioles. Hence, there is some reversion by the centrosomes of cultured cells to the operational configuration regarded as typical for animal tissue cells in general.

Taxol influences control of protofilament number at microtubule-nucleating sites in Drosophila.

Control of protofilament number has been investigated using Drosophila wings at a stage when 15-protofilament microtubules assemble under normal conditions. Microtubule nucleation still progressed at the usual microtubule-nucleating sites in the presence of taxol. However, provided taxol was introduced before microtubule nucleation began, few microtubules with 15 protofilaments assembled. Most microtubules were composed of 12 protofilaments (a previously undetected value for Drosophila) or 13 protofilaments (which is the value for microtubules in most eukaryotic cells). Unexpectedly, a comparatively mild challenge to control of nucleation (in vitro wing culture) also promoted assembly of 13-protofilament microtubules. Hence, the microtubule-nucleating sites may possess a relatively labile control specifying 15 protofilaments superimposed upon that for maintaining 13-protofilament fidelity.

Microtubule polarities indicate that nucleation and capture of microtubules occurs at cell surfaces in Drosophila.

Hook decoration with pig brain tubulin was used to assess the polarity of microtubules which mainly have 15 protofilaments in the transcellular bundles of late pupal Drosophila wing epidermal cells. The microtubules make end-on contact with cell surfaces. Most microtubules in each bundle exhibited a uniform polarity. They were oriented with their minus ends associated with their hemidesmosomal anchorage points at the apical cuticle-secreting surfaces of the cells. Plus ends were directed towards, and were sometimes connected to, basal attachment desmosomes at the opposite ends of the cells. The orientation of microtubules at cell apices, with minus ends directed towards the cell surface, is opposite to the polarity anticipated for microtubules which have elongated centrifugally from centrosomes. It is consistent, however, with evidence that microtubule assembly is nucleated by plasma membrane-associated sites at the apical surfaces of the cells (Mogensen, M. M., and J. B. Tucker. 1987. J. Cell Sci. 88:95-107) after these cells have lost their centriole-containing, centrosomal, microtubule-organizing centers (Tucker, J. B., M. J. Milner, D. A. Currie, J. W. Muir, D. A. Forrest, and M.-J. Spencer. 1986. Eur. J. Cell Biol. 41:279-289). Our findings indicate that the plus ends of many of these apically nucleated microtubules are captured by the basal desmosomes. Hence, the situation may be analogous to the polar-nucleation/chromosomal-capture scheme for kinetochore microtubule assembly in mitotic and meiotic spindles. The cell surface-associated nucleation-elongation-capture mechanism proposed here may also apply during assembly of transcellular microtubule arrays in certain other animal tissue cell types.

Intermicrotubular actin filaments in the transalar cytoskeletal arrays of Drosophila.

Rabbit muscle myosin subfragment S1 decorates 6 nm diameter filaments in Drosophila wing epidermal cells in the arrowhead fashion characteristic of the binding of subfragment S1 to actin filaments. The filaments in question are concentrated between microtubules that are mostly composed of 15 protofilaments and form cell surface-associated transcellular bundles. There are indications that the majority of the actin filaments have the same polarity and that, like the microtubules, they may elongate from sites at the apical surfaces of the cells. The bundles of F actin and microtubules occur in dorsal and ventral epidermal cell layers of a wing blade. They are joined in dorso-ventral pairs by attachment desmosomes. These transalar cytoskeletal arrays may provide an example of a situation where actin filaments operate as stiffeners rather than active generators of force in conjunction with myosin. The arrays probably function as noncontractile pillars to maintain basal cell extensions and keep haemocoelic spaces open in the highly folded and expanding wing blades of late pupae.

Evidence for microtubule nucleation at plasma membrane-associated sites in Drosophila.

This report is concerned with the nucleation and organization of microtubule bundles that assemble after 'conventional' centrosomal microtubule-organizing centres have been lost. The microtubule bundles in question span the lengths of wing epidermal cells. Bundles extend between hemidesmosomes at the apical cuticle-secreting surfaces of cells and basal attachment desmosomes that unite the dorsal and ventral epidermal layers of developing wing blades. Furthermore, each bundle includes up to 1500 microtubules and most of the microtubules are composed of 15 protofilaments. Individual cells were serially cross-sectioned at an early stage of bundle assembly. The number of microtubule profiles/cell cross-section decreased progressively by up to 59% of the most apical values in section sequences cut from fairly apical to more basal levels in the cells. The apical ends of microtubules were associated with numerous small dense plaque-like sites (diameter 0.1-0.2 micron), which were specialized regions of plasma membranes at the apical surfaces of cells. Many of the microtubules near apical plaques were not well aligned with each other; they 'radiated away' from cell apices. This was in contrast to the situation at more basal levels where most microtubules were oriented parallel to the longitudinal axes of cells. These findings indicate that the relatively dispersed arrays of apical plasma membrane-associated plaques act as microtubule-nucleating sites to initiate basally directed elongation of bundle microtubules. Apical cell surfaces and their plaques seem to operate as microtubule-nucleating and -organizing regions that functionally replace the centrosomal microtubule-organizing centres lost earlier in cell differentiation.